Method and apparatus for acquiring feature data from low-bit image

A processor-implemented method of generating feature data, the method comprising:

• receiving a feature map of an input image including pixels having a plurality of bit value levels;
• generating, for each of the bit value levels, a binary feature map including binary pixe l values corresponding to pixels of the feature map that include a bit of the bit value level;
• and
• generating, using a neural network, feature data corresponding to the input image based on the generated binary feature maps,
• wherein the generating the binary feature map comprises:
• generating M color channel images in response to the input image being a color image including M color channels, wherein each of the M color channel images corresponds to each of the M color channels of the input image and M is an integer equal to or greater than 2;
• and
• generating the binary feature map from the M color channel images.

An apparatus for generating feature data, the apparatus comprising:

• one or more processors configured to:
• receive an input image;
• generate N binary images from the input image, wherein N is the number of bits per pixel of the input image, wherein N is an integer equal to or greater than 2;
• and
• generate, using at least one neural network, feature data corresponding to the input image from the N binary images,
• wherein for the generating N binary images from the input image, the one or more processors are configured to:
• generate M color channel images in response to the input image being a color image including M color channels, wherein each of the M color channel images corresponds to each of the M color channels of the input image and M is an integer equal to or greater than 2;
• and
• generate N binary images from the M color channel images.